Although Drug Court (DC) programs have been in place for over 15 years the research emphasis has been on recidivism outcomes after graduation from the program, and service factors relating to graduation and recidivism as outcomes. This study focuses on using the Addiction Severity Index (ASI) as an assessment tool in DC programs, and analyzes the predictive relationship between ASI composite scores and selected item responses and DC program compliance (urinalyses testing positive for drug use) and graduation. Data from 113 participants who either graduated (n = 76, 67%) or were removed from (n = 37, 33%) an adult Drug Court program in a medium size Southwestern community were used to answer research questions that included; (1) What is the relationship between ASI composite scores and DC program compliance? (2) What is the relationship between ASI composite scores and graduating from a DC program? (3) What is the relationship between selected ASI item responses and DC program compliance? and (4) What is the relationship between selected ASI item responses and graduating from a DC program? The selected ASI items included those focused on lifetime history of chemical dependency symptoms, lifetime previous alcohol or other drug (AOD) treatment, and recent use of cocaine, stimulants, narcotics, and alcohol. The seven ASI composite scores focus on medical, employment, alcohol, drug, legal, family, and psychiatric problems. A series of four Logistic Regression analyses were used to answer the research questions. Results revealed that; (1) Higher alcohol composite scores significantly predicted greater DC program compliance, and higher drug, legal, and psychiatric composite scores were associated with lower compliance, (2) Higher drug composite scores significantly predicted lower odds of graduating from DC, and higher psychiatric composite scores were associated with greater odds of graduating, and (3) That although selected item responses failed to effectively predict DC program compliance, a lifetime history of withdrawal symptoms, prior drug rehabilitation participation, and use of stimulants within 30 days of beginning DC participation all significantly predicted DC graduation. All Logistic Regression models were evaluated for their improvement in prediction over models with the constant alone, explained variance, classification sensitivity and selectivity, and goodness of fit with an ideal model. All models performed satisfactorily in terms of improvement in prediction and goodness of fit, but explained variance (between 10% and 30%) and classification sensitivity and selectivity was less consistent across models. These findings warrant replicating this line of inquiry with larger samples drawn from multiple DC programs. Practice implications include the use of ASI composite scores and item responses to tailor service planning for subsets of individuals within the structure of DC programs. In addition to greater intensity, and duration of intensive services, these findings speak to the possibility of individualizing treatment services in the DC context while remaining systematic through using data for such DC practice innovations.